Identifying factors for student retention of higher ed institutions using decision tree

Abstract

Student retention is an issue of high priority for many colleges and universities. Keeping students in school is the very basic condition for them to achieve their goals for going to colleges in the first place. A lot of research and practices have been done across institutions to improve student retention rates, but colleges and universities are still trying to figure out what are the factors that are most important to student retention. In this paper, we present our experiments of building predictive models, particularly decision tree models, to fit in the overall prediction of full time student retention. The data set of 1,965 cases from 1987 to 2000 obtained from the Delta Cost Project Database of the American Institutes for Research has 541 variables. We used variable selection measures like R-Squared to reduce to 45 variables and build decision tree models to fit the training data. Eight variables were identified to be most influential to the retention rates. Our experiments show that the decision trees with moderate depth are suitable for creating retention model.